Couples’ money arrangements in Germany: A visualization of cohort, age group, and partnership type variations

Supplementary Material

Authors

Marcel Raab

Florian Schulz

Date

Oct 31, 2024

This document contains supplementary material for our visualization of couples’ money arrangements in Germany. The code files used for editing the source data and generating the visualization as well as this document (including the tables) are available on GitHub.

The visualization and this supplement build on previous work of Pepin (2022) and are, therefore, similarly structured. While Pepin has focused on US data and differences in money arrangements by marital and parental status, by age group, and for different-gender and same-gender couples, we utilized the unique cohort-sequence design of the German Family Panel (“pairfam”) that allows for between and within cohort comparisons.


Data

The visualization and the supplementary tables are based on data from the German Family Panel (pairfam), release 13.0 (Brüderl, Drobnič, et al. 2022). The annually collected survey data from a nationwide random sample comprised data from the three birth cohorts 1971-73, 1981-83, and 1991-93. We used data from waves 1 and 9, collected in 2008/09 and 2016/2017. We applied calibrated design weights adjusting the samples to central characteristics of the German population for the three initial cohorts. A detailed description of the study can be found in Huinink et al. (2011).


Sample construction

Our analytical approach took advantage of pairfam’s cohort-sequence design that allows for comparing distributions of variables measured for different cohorts at the same age and for members of the same cohort at different ages. Specifically, our analysis included two cohort comparisons of couples’ money arrangements: the first compares persons aged 34 to 36 from the birth cohort 1971-1973 and the birth cohort 1981-1983; the second comprised persons aged 24 to 26 from the birth cohort 1991-1993 and the birth cohort 1981-1983.

The sample construction aimed at comparing like with like; hence, we only kept persons from overlapping age ranges (e.g., persons from cohort 1981-1983 aged 32 to 33 were not considered because we did not observe equally aged persons in cohort 1971-1973).

As the birth cohort 1981-1983 contributed twice, we also show the respective within-cohort comparison for two age groups mentioned above. Survey data for these comparisons came from pairfam waves 1 and 9, collected in 2008/09 and 2016/17. Our starting sample (with overlapping age groups) included 7,539 cases.

We restricted this baseline sample to respondents living in a partnership by deleting 32 cases with a missing value on partnership status and 1,984 respondents who reported being single. Finally, we deleted another 167 cases with missing information on the couples’ financial arrangements.

This yielded a final sample size of 5,523. As members of the middle birth cohort could contribute twice (wave 1 and 9), this number does not represent the number of distinct respondents. In fact, 253 respondents appeared twice in the final data. That is, 2,686 observations for this cohort came from 2,433 respondents. Cohort 1971-1973 contributed 2,066 cases; the youngest cohort (1991-1993) constituted the smallest group (n = 604).


Measurement of couples’ money arrangements

We followed the categorization of Pepin (2022)(1) We keep all of our money separately, (2) Put some money together, or (3) Put all of our money together—and applied it to the pairfam question shown in Figure 1. As already mentioned, only very few cases did not respond to this question (167 of 5,523 cases, corresponding to 3.0% of the analysis sample).

Figure 1: Snapshot from pairfam (wave 1) questionnaire pairfam Group (2022:126)

We implemented Pepin’s (2022) coding scheme by applying following categorization:

  • Put all money together
    • If the respondent exclusively mentioned “inc19i1”
  • Put some money together
    • If the respondent mentioned “inc19i1” and at least one other option
    • If inc19i2 or inc19i3 were mentioned (irrespective of which other options have been chosen)
  • Keep all money separate
    • If the respondent mentioned “inc19i4” and/or “inc19i5” and neither of the other three options


Distribution of partnership states

Cohort and age group differences in couples’ money arrangements were considerably affected by group-specific differences in the distribution of partnership states, as shown in Table 1. Accordingly, our visualization shows couples’ money arrangements for entire cohorts or age groups and separately by partnership states. The statistical tests for group differences displayed in the next section proceeded accordingly.

Table 1
Distribution of partnership states across cohorts and age groups
Table shows weighted age-specific column percentages and case numbers. Unweighted case numbers in brackets.
Partnership status
1971-1973
1981-1983
1991-1993
Age 34–36 Age 24–26 Age 34–36 Age 24–26
    Living apart together 9%
191 [161]
36%
648 [612]
10%
85 [78]
39%
232 [255]
    Cohabitation 18%
394 [355]
37%
673 [705]
23%
203 [221]
43%
259 [279]
    Marriage 74%
1,630 [1,550]
27%
495 [505]
67%
580 [565]
18%
106 [70]


Examining group differences with \({\chi}^2\)-tests

We computed design-adjusted Rao‐Scott \({\chi}^2\)-tests to test for statistical differences within and between cohorts. These tests use survey calibrated design weights aiming “to adjust the data to the target population and simultaneously control for baseline survey participation and panel attrition bias” (Brüderl, Garrett, et al. 2022:63). The tables below display unweighted and weighted frequencies to provide information about the cell sizes and row percentages to ease the comparisons across different sub groups. For both between and within cohort comparisons, we analyzed group differences for the total sample and by partnership status.


Test of cohort differences

The tests displayed in Table 2 suggested that couples’ money arrangements were not equally distributed across cohorts. A fine-grained analysis by partnership status indicated that this result was mainly driven by the notable cohort differences among married respondents.

Table 2
Cohort differences in couples’ money arrangements across two age groups
Table shows weighted row percentages by cohort and case numbers. Unweighted case numbers in brackets. p-values from design-adjusted Rao‐Scott chi-square tests.
Age 24–26
Age 34–36
Put all money together Put some money together Keep all money separate p-value Put all money together Put some money together Keep all money separate p-value
Total


<0.001


<0.001
        1991-1993 8%
47 [37]
24%
146 [139]
68%
404 [428]





        1981-1983 19%
340 [369]
22%
395 [419]
59%
1,080 [1,034]

28%
239 [234]
36%
313 [308]
36%
316 [322]

        1971-1973



42%
923 [887]
30%
673 [635]
28%
620 [544]

Living apart together


0.9


0.8
        1991-1993 1%
2 [3]
5%
12 [13]
94%
218 [239]





        1981-1983 1%
7 [9]
6%
38 [37]
93%
602 [566]

0%
0 [0]
12%
10 [7]
88%
75 [71]

        1971-1973



1%
1 [1]
10%
19 [18]
89%
171 [142]

Cohabitation


0.052


0.3
        1991-1993 5%
12 [15]
35%
91 [95]
60%
155 [169]





        1981-1983 10%
66 [87]
31%
207 [225]
60%
400 [393]

8%
17 [21]
32%
66 [73]
59%
120 [127]

        1971-1973



14%
55 [51]
32%
125 [120]
54%
214 [184]

Marriage


0.004


<0.001
        1991-1993 31%
33 [19]
40%
43 [31]
29%
31 [20]





        1981-1983 54%
267 [273]
30%
151 [157]
16%
77 [75]

38%
222 [213]
41%
237 [228]
21%
121 [124]

        1971-1973



53%
866 [835]
32%
529 [497]
14%
235 [218]


Test of age group differences within cohort 1981-1983

Table 3 shows that couples’ money arrangements among members of the birth cohort 1981-1983 differed across age. Breaking down the comparisons by partnership status revealed that the age groups only differed significantly among the married respondents. Younger married persons were less likely to pool their money entirely.

Table 3
Age group differences in couples’ money arrangements for the birth cohort 1981-1983
Table shows weighted row percentages by age group and case numbers. Unweighted case numbers in brackets. p-values from design-adjusted Rao‐Scott chi-square tests.
Characteristic Put all money together Put some money together Keep all money separate p-value
Total


<0.001
        24–26 19%
340 [369]
22%
395 [419]
59%
1,080 [1,034]

        34–36 28%
239 [234]
36%
313 [308]
36%
316 [322]

Living apart together


0.2
        24–26 1%
7 [9]
6%
38 [37]
93%
602 [566]

        34–36 0%
0 [0]
12%
10 [7]
88%
75 [71]

Cohabitation


0.9
        24–26 10%
66 [87]
31%
207 [225]
60%
400 [393]

        34–36 8%
17 [21]
32%
66 [73]
59%
120 [127]

Marriage


<0.001
        24–26 54%
267 [273]
30%
151 [157]
16%
77 [75]

        34–36 38%
222 [213]
41%
237 [228]
21%
121 [124]


Parental status

Due to sample size limitations, the main document focuses on differences in couples’ money arrangements by partnership status. Figure 2 presents a more detailed analysis that additionally differentiates by parental status. Given the small sample sizes and skewed distributions, the analysis was restricted to individuals who are cohabiting or married and who are from the two older birth cohorts. For the married respondents the results were consistent with those shown in Figure 1 of the main article. While there were level differences between parents and childless respondents, both groups showed a decline in joint pooling across cohorts and age groups. A similar pattern was observed among parents in co-residential unions.


Data availability statement

The data of the German Family Panel (pairfam) is available for scientific purposes from GESIS Leibniz Institute for the Social Sciences: https://doi.org/10.4232/pairfam.5678.13.0.0


References

Brüderl, Josef, Sonja Drobnič, Karsten Hank, Franz. J. Neyer, Sabine Walper, Philipp Alt, Elisabeth Borschel, Christiane Bozoyan, Madison Garrett, Svenja Geissler, Tita Gonzalez Avilés, Nicolai Gröpler, Kristin Hajek, Michel Herzig, Rüdiger Lenke, Renate Lorenz, Katharina Lutz, Timo Peter, Richard Preetz, Julia Reim, Barbara Sawatzki, Claudia Schmiedeberg, Philipp Schütze, Nina Schumann, Carolin Thönnissen, Katharina Timmermann, and Martin Wetzel. 2022. The German Family Panel (Pairfam).”
Brüderl, Josef, Madison Garrett, Kristin Hajek, Michel Herzig, Rüdiger Lenke, Renate Lorenz, Philipp Schütze, Nina Schumann, and Katharina Timmermann. 2022. “Pairfam Data Manual, Release 13.0. LMU Munich: Technical Report. GESIS Data Archive, Cologne. ZA5678 Data File Version 13.0.0.” doi: 10.4232/pairfam.5678.13.0.0.
Huinink, Johannes, Josef Brüderl, Bernhard Nauck, Sabine Walper, Laura Castiglioni, and Michael Feldhaus. 2011. “Panel Analysis of Intimate Relationships and Family Dynamics (Pairfam): Conceptual Framework and Design.” Zeitschrift Für Familienforschung 23(1):77–101. doi: 10.20377/jfr-235.
pairfam Group. 2022. “Anchor Codebook, Wave 1 (2008/2009), Release 13.0. GESIS Data Archive, Cologne. ZA5678.”
Pepin, Joanna R. 2022. “A Visualization of U.S. Couples Money Arrangements.” Socius: Sociological Research for a Dynamic World 8. doi: 10.1177/23780231221138719.